Commit 3c5745a2 authored by Hines, Jesse's avatar Hines, Jesse
Browse files

More updates to support different systems

parent 4e64224c

raps @ 14f93d32

Compare ddcdeb5d to 14f93d32
Original line number Diff line number Diff line 
Subproject commit ddcdeb5d29fa0a321bef4800e0e93a42a4e6293f
 
Subproject commit 14f93d327ab837999461e7173dff012125497d19

simulation_server/simulation/simulation.py

+7 −17
Original line number Diff line number Diff line
@@ -12,7 +12,6 @@ from .raps.raps.power import PowerManager 
from .raps.raps.flops import FLOPSManager

from .raps.raps.scheduler import Scheduler

from .raps.raps.telemetry import Telemetry

from .raps.raps.dataloaders.frontier import index_to_xname

from .raps.raps.workload import Workload

from ..models.sim import SimConfig, SimSystem

from ..models.output import (
@@ -37,16 +36,6 @@ def _offset_to_time(start, offset): 
        return None





CDU_INDEXES = [

    'x2002c1', 'x2003c1', 'x2006c1', 'x2009c1', 'x2102c1', 'x2103c1', 'x2106c1', 'x2109c1',

    'x2202c1', 'x2203c1', 'x2206c1', 'x2209c1', 'x2302c1', 'x2303c1', 'x2306c1', 'x2309c1',

    'x2402c1', 'x2403c1', 'x2406c1', 'x2409c1', 'x2502c1', 'x2503c1', 'x2506c1', 'x2509c1',

    'x2609c1',

]

def _cdu_index_to_name(index: int):

    return CDU_INDEXES[index - 1]





class SimOutput(NamedTuple):

    timestamp: datetime

    scheduler_sim_system: list[SchedulerSimSystem]
@@ -141,7 +130,6 @@ def get_scheduler( 
    return Scheduler(

        power_manager = power_manager,

        flops_manager = flops_manager,

        layout_manager = None,

        cooling_model = cooling_model,

        debug = False, replay = replay,

        schedule = schedule_policy,
@@ -178,12 +166,13 @@ def run_simulation(sim_config: SimConfig): 
            replay = (sim_config.scheduler.jobs_mode == "replay"),

            schedule_policy = sim_config.scheduler.schedule_policy,

        )

        telemetry = Telemetry(system = sim_config.system, config = sc.config)



        # Memoized function to convert raps indexes into node names.

        # Memo increases performance since it gets called on snapshots of the same job multiple times.

        @functools.lru_cache(maxsize = 65_536)

        def _parse_nodes(node_indexes: tuple[int]):

            return [index_to_xname(i, sc.config) for i in node_indexes]

            return [telemetry.node_index_to_name(i) for i in node_indexes]



        if sim_config.scheduler.jobs_mode == "random":

            num_jobs = sim_config.scheduler.num_jobs if sim_config.scheduler.num_jobs is not None else 1000
@@ -240,7 +229,8 @@ def run_simulation(sim_config: SimConfig): 


            scheduler_sim_jobs: list[SchedulerSimJob] = []

            curr_jobs = {}

            for job in data.jobs:

            data_jobs = data.completed + data.running + data.queue

            for job in data_jobs:

                # end_time is set to its planned end once its scheduled. Set it to None for unfinished jobs here

                if job.start_time is not None:

                    time_end = _offset_to_time(sim_config.start, job.end_time)
@@ -270,7 +260,7 @@ def run_simulation(sim_config: SimConfig): 
            prev_jobs = curr_jobs



            power_history: list[SchedulerSimJobPowerHistory] = []

            for job in data.jobs:

            for job in data_jobs:

                if job.id and power_history_counts.get(job.id, 0) < len(job.power_history):

                    power_history.append(SchedulerSimJobPowerHistory(

                        timestamp = timestamp,
@@ -352,8 +342,8 @@ def run_simulation(sim_config: SimConfig): 




            for cdu_index, cdu_data in cooling_sim_cdu_map.items():

                cdu_name = _cdu_index_to_name(cdu_index)

                row, col = int(cdu_name[2]), int(cdu_name[3:5])

                cdu_name = telemetry.cdu_index_to_name(cdu_index)

                row, col = telemetry.cdu_pos(cdu_index)

                cdu_data.update(

                    timestamp = timestamp,

                    name = cdu_name,